Six years ago, a catastrophic flood unfolded in the Lockyer Valley in southeast Queensland. The floodwaters then spread on January 11 2011 across the Brisbane River floodplains, inundating the Brisbane CBD and inner suburbs, and bringing the state’s capital to a standstill.
The January floods came in the wake of other devastating floods that had hit Queensland towns and cities in December 2010, affecting an area bigger than France and Germany combined. Thirty-three people died in the 2010/2011 floods; three remain missing.
Our research, based on palaeological flood records, suggests floods of the size of January 2011 may be more common than we think. When the next one occurs is a matter of when, not if. So what can we do to plan better?
What is a one-in-100-year flood?
Floods are the most expensive type of natural disaster in Australia. The 2011 flood is estimated to have cost the Australian economy around A$30 billion. This does not include the incalculable cost of declining water quality and ecosystem health in offshore ecosystems such as Moreton Bay.
To manage flood risk, we have to understand the chance of different-sized floods occurring. The chance of a flood event can be described using a variety of terms, commonly including the average recurrence interval (ARI). You’ve probably seen this reported in the media as the “one-in-100-year flood”.
However, the preferred method is now annual exceedance probability (AEP). For example, the one-in-100-year flood has a one-in-100 chance or 1% AEP of being exceeded in any year. Currently, this 1% AEP event is designated as having an “acceptable” risk for planning purposes nearly everywhere in Australia.
Initial estimations of the 2011 event, based on 31 years of gauging records in the Upper Lockyer, indicated an AEP of 0.05%, or an ARI of one in 2,000 years.
But another extreme event in 2013, with five more years of data, reduced this to 1.11%, or one in 90 years. This illustrates a major problem with calculating flood risk: flood prediction is extremely dependent on the amount of data. It is worse in countries such as Australia where European settlement occurred relatively recently.
Finding old floods
One way to extend the data is to incorporate palaeoflood records into flood predictions. Palaeoflood records are obtained from a range of techniques that combine different sources of past flood information from the landscape.
These might include markers on old buildings or bridges which extend further back than the river gauging records. Flood sediments stored high in bedrock gorges or in lowland floodplains also provide a long-term record once we’ve dated them.
In our recent Australian Research Council Linkage Project, The Big Flood: will it happen again?, we’ve been looking at these types of flood records.
The project has produced the first-ever palaeoflood record for the Lockyer Valley, extending it back several thousand years. We’ve produced a timeline of past floods using a technique called Optically Stimulated Luminescence, which estimates the age based on how much sunlight (UV light) is stored in a single grain of sand. UV light produces a luminescence signal that gets trapped inside the lattice of the quartz sand. The amount stored can be converted to produce an age since burial.
This record reveals that flood events like 2011 have occurred at least seven times over the past 1,000 years. We also found a period of high flood activity during the 1700s, which exceeded the size of the historical events of the 1890s and 1974 floods. We can also see clusters of floods within short time periods, such as the cluster in the 1800s, which was highlighted again by the floods in 2011 and 2013.
The record indicates that such extreme flood events may occur more frequently than we thought.
Most importantly, when we incorporate palaeoflood records into traditional flood analysis, the uncertainty in predictions is significantly reduced.
What do we need to do with this information?
Palaeoflood records represent a viable, cost-effective solution to the ongoing problem of flood risk management in Australia. To date, the use of palaeoflood records has not been included in traditional flood analysis nor recognised in planning or policy. In spite of two extreme events in 2011 and 2013, many planning and policy guidelines remain unchanged.
The two extreme flood events in 2011 and 2013 indicate that the level of certainty around acceptable limits of flooding is inadequate. Longer records are needed to reduce flood risk in southeast Queensland. Without this critical next step, Australians remain at risk of extreme flood events.
Given the likelihood of increasing rainfall extremes in the future, it is important we start using the information nature has preserved to better prepare for more frequent extreme floods.
* Correction: The original version of this story stated that the flooding of the Brisbane floodplains in January 2011 was equivalent to the area of France and Germany combined. That was incorrect; it should have referred to the total area affected by floods across Queensland in the 2010/11 floods. This article has now been updated and corrected with details from the 2012 Queensland Floods Commission of Inquiry. Thank you to readers David Arthur and Mark Duffett for pointing out the error.